33. Problem Formulation. Problem formulation, i.e. formulation of risk hypotheses requires precise definition of relevant sources and targets of suspected harm. Under-utilisation of this tool may lead to poor risk assessment. Furthermore, for nanomaterials, the mostly limited depth of information (qualitative and quantitative) on sources and targets of harm may represent a hurdle in problem-formulation. The SG6 2009 Workshop (OECD 2010a) included a discussion on problem formulation needs:
• Consider the “particle nature” of the material, such as the surface properties and interactions, the relation of metrics used, the characteristics of the material;
• Assess and accommodate risk assessment approaches with regard to the effects of test methods and exposure matrix (e.g., dispersion methods) on testing outcomes and on inter-comparability of the data used in the assessment; and
• Include particular attention to the complex nature of the material (e.g., variation in size, surface properties, and composition that create a heterogeneous range of particle types) and its interaction with environmental components and transport mechanisms in exposure and toxicity contexts.
34. Sources of Potential Harm. Nanoparticles are known to be unintentionally produced and released into the atmosphere by natural phenomena and many human industrial and domestic endeavours, such as transportation utilizing internal combustion and jet engines,. In recent years a new type of source of nanoparticles has been introduced, within the sphere of intentionally engineered nanoscale components of consumer products and advanced technologies. For these engineered nanoscale components, two separate types of nanostructure may be identified, those where the structure itself is a free particle (or agglomerate or aggregate thereof), and those where the nanostructure is an integral feature of a larger object (e.g. an ultrathin surface coating or semiconducting layer). Particularly the many uncertainties in production, use and fate and behaviour of free insoluble nanoparticles give rise to concerns over possible human health and environmental risks. However, SCENIHR (2009) also concludes that “The hypothesis that smaller means more reactive and thus more toxic cannot be substantiated by the published data. In this respect nanomaterials are similar to normal substances in that some may be toxic and some may not. As there is not yet a generally applicable paradigm for nanomaterial hazard identification, a case by case approach for the risk assessment of nanomaterials is recommended.” Nevertheless, information on mode of action and structure-activity relationships may facilitate development of categorical-based hazard and risk characterisation (OECD 2007).
35. Nanomaterial Identification. The nanomaterial for which a risk assessment is performed (i.e.
the scope) needs proper definition. In the absence of an international definition, OECD WPMN has applied for the practical work a working definition based on ISO and other relevant considerations. The identification of a nanomaterial includes appropriate naming as a key element. Properties to be considered as identifier could include chemical composition, crystallinity, surface coatings, morphology, size (range), etc. (OECD 2009c).
36. Variability in Composition and Properties. Unlike discrete chemicals, nanomaterials can be present as substances with a variable composition that goes beyond variations at the level of impurities.
Examples include variations in size and size distribution, surface properties or composition of the nanoform itself. Strategies to accommodate for this particular character also during testing should be developed and introduced.
37. Nanomaterial Characterisation. In the context of nanomaterial characterisation it is noted that describing the properties of the primary nano-objects of the material (nanoparticles, -fibres, -sheets) themselves is essential but not sufficient. The interactions between the nanoobjects within a given environment or formulation must be considered as well as interactions of the material with components thereof. The physicochemical properties and material characterisation that may be required for testing are described in more detail in OECD (2008). Specialised instrumentation that is not usually available in test facilities may also be needed for characterisation of the material within the vehicle (as prepared for dosing).
38. Selection of Assessment Endpoints. To provide direction and boundaries for RA, the specific entity to be protected, such as individuals, a species, a sub-population, a community, an ecosystem, etc. has to be identified. In addition, the concerns or effects to be protected from (e.g. reduced survival and reproductive impairments in ecological RA) generally require definition. For nanomaterials, this may be complicated by an incomplete knowledge about its behaviour throughout the lifecycle and limited experience with toxicity in the target species or population.
39. Testing Plan. Generally, good problem formulation allows for clear definition of the minimum data required to show safety. For regulation of conventional chemicals, standard data requirements have been prescribed, based on extensive experience, for substance categories such as pesticides. Such standard requirements may need adaptation for nano-scale materials. The same applies to the practice of use of other existing data and methods to “bridge” to that existing data. Especially the level of generalisation that can and should occur needs to be evaluated and a scientifically sound approach that allows for inclusion of information obtained using dissimilar materials (even non-nanoscale), methods or reporting has to be defined (OECD 2007).
3.b. Hazard Identification
40. Applicability of Testing Methods. The direct hazard that specific nanomaterials present to human health and to the environment will depend on the physicochemical (and chemical) characteristics of the surface and core of the nano-objects, and the extent to which the material exhibits interactions with biological systems associated. SCENIHR (2006) noted the insufficiency of scientific information about the physiological responses to nanoparticles, about the mechanisms of interaction at sub-cellular level (see also below), and about the changes in the nanoparticle physicochemical characteristics like agglomeration and aggregation, surface modification, dissociation, degradation, adsorption of different species, etc. Those changes would depend on the size/shape of the particle as well as on the local environmental and cellular conditions (ionic strength, acidity, viscosity, etc.) Therefore, the methods used in the hazard identification and assessment may also need to be augmented to include all of the above considerations.
41. Endpoints Assessed. Hazards are commonly identified in standardised acute and chronic (eco) toxicity tests. As concluded by SG4 in its Preliminary Review of OECD Test Guideline for their Applicability to Manufactured Nanomaterials (OECD 2009a,), the OECD guidelines are in general considered applicable to manufactured nanomaterials, particularly with regard to investigating their health effects, with the important proviso that additional consideration needs to be given to the physicochemical characteristics of the material tested, including dosing. In some cases, there may be a need for further modification to the OECD guidelines. Preparation of samples and dose administration are critical considerations for the tests and therefore guidance has been developed on sample preparation and dosimetry for the safety testing of nanomaterials (OECD 2010c). The preliminary review of OECD-WPMN is consequently seen as a “living” document, highlighting the feasibility of various approaches and allowing for continuous updates, given the rapid developments in this area. Nevertheless, there may be remaining uncertainty at the moment in the respect that specific toxicity (mechanisms) related to the size or the particle nature of specific nanomaterial may be overlooked since standardised tests are usually aimed at
a specific endpoint based on experience with, supposedly, non-nanomaterials. As the field of nanotoxicology advances, this uncertainty may be reduced through additional research.
42. Target Organs. The current OECD Test Guidelines in principle enable the assessment of the all the possible target organs affected (OECD 2009a). Toxicokinetic studies may provide useful information in this context. More specifically, given the indications that nanoparticles could migrate from the respiratory tract to the blood and on to the brain (or translocate directly via the olfactory nerves), SCENIHR (2006) emphasised the need for the development of quantitative assays that could determine the presence of actual nanoparticles in different tissues of the human or animal body. To date, toxicokinetics usually relies on measurement of the primary matter or bound residues of metal catalysts rather than the nanomaterial as such (e.g. Ti for nano-TiO2 or Co for Baytube CNTs; Chen 2009, Pauluhn 2010a). Taking into account the slow body clearance observed for some nanomaterials (e.g. Chen 2009, Pauluhn 2010a), local accumulation may play an important role.
43. Effective Dose. The effective concentration or dose (that results in an adverse biological response) derived for a manufactured nanoparticle from laboratory studies is likely to be influenced by the abiotic (and biotic) composition of the exposure pathway, variations in which may influence nanoparticle structure, form and behaviour.. To give one example, in aquatic systems some relevant abiotic factors are pH, ionic strength and the concentration of humic substances in the aquatic matrix. These are known to influence and modify physicochemical characteristics of the particle, notably agglomeration and aggregation. Notably, the effective dose of a nanomaterial may be smaller on a mass basis than the effective dose of larger particles of the same material if the mode of action relates to the total particle number or surface area (Handy 2008).
44. External Factors Influencing Toxicity. One factor determining particle behaviour is how the particular natural environment will influence important physicochemical characteristics such as surface charge, and/or agglomeration and aggregation. Thus, abiotic factors may play critical roles in the context of bioavailability, distribution, bioaccumulation and, ultimately, toxicity of nanomaterials when exposure occurs in natural settings. In some cases, specific environmental components, esp. biopolymers, absorb stably to a particle surface (Handy 2008). One phenomenon identified in this context is the formation of a
“protein corona” (Maiorano 2010).
45. Variability of External Factors. A number of abiotic and biotic factors that influence nanoparticle toxicity may be variable themselves as well, depending on the (receiving) environment, which can be highly complex (e.g. estuaries where pH and ionic strength can vary considerably) (Handy 2008). In principle, this is an issue not exclusive to nanomaterials, but the specific factors of relevance, their variability and impact may be different from what is expected.
46. Definition of Adversity. OECD Test Guidelines refer to adverse effects and define it in a following manner: “Change in the morphology, physiology, growth, development, reproduction or life span of an organism, system, or (sub) population that results in an impairment of functional capacity, an impairment of the capacity to compensate for additional stress, or an increase in susceptibility to other influences” (OECD 2003). For nanomaterials, there is debate about the definition of adversity for specific effects. One example is, whether the presence of nanoparticles in the brain is an adverse event as such or if there should be (indication of) proof that the brain function or structure is negatively affected by the presence of the nanoparticles before it can be regarded as an adverse effect. This applies to human health hazard assessments as well as to ecotoxicology.
47. Mechanistic Considerations. A number of mechanisms by which toxic nanoparticles may exert their effects have been proposed and these are summarised in the figure below (adapted from SCENIHR
2006). However, a full understanding of the various mechanisms involved in nanomaterial toxicity is still
48. Non-Nano – to – Nano Extrapolation. As already pointed out by other organisations and bodies (e.g. SCENIHR 2010), the term “nanomaterial” relates to a categorisation by size, and would not per se imply specific risks or hazard properties. It is noted, however, that reduction in size to the nanoscale may or may not change the characteristics of particles (SCENIHR 2009), and observed differences in biological effects may be due to the increased surface to volume or surface to mass ratio. In addition, size can influence the materials (bio) distribution and the kinetics thereof in an organism or an ecosystem.
Finally, the structuring of matter at the nanoscale is characterised by the interplay of classical physics and quantum mechanics which could lead to some novel characteristics vis-à-vis the same material without nanoscale features. It was apparent from discussion during the SG6 workshop that the development of relationships between existing data on nanoscale and non-nanoscale materials may be difficult due to limited or lack of data (OECD 2010a). For some materials (e.g., poorly soluble low toxicity particles), the surface area of the particles has been related to the lung response, such that nanoparticles were more inflammogenic than the same mass of larger particles of the same chemical composition (Bermudez et al.
2002, 2004; Elder et al. 2005). In these cases, hazard/risk grouping strategies may be considered for particles with the same mode of action.
49. Nano – to – Nano Extrapolation. The nanomaterial properties determining its toxicokinetics and toxicodynamics are currently not known with confidence. Therefore, it should be realised that different nanoforms/sizes may show differences in effect concentrations and/or effect parameters. Importantly, these uncertainties mean that methodologies to permit extrapolation between different types of nanomaterials and different species are not available, implying that assessments often have to be made de novo, on a case-by-case basis (see also “problem formulation”). Additional data are needed to link the biological effects with the physicochemical properties of nanomaterials in order to develop predictive hazard/risk grouping strategies (Rushton et al. 2010, OECD 2010b, OECD 2010c).
50. Nanomaterials Acting as Carriers. Chemical risk assessments may consider whether chemicals absorb onto particulate matter as this phenomenon can influence the transport and compartmentalisation of chemicals (SCENIHR 2006). Likewise, the particulate nature of nanomaterials may result in the adsorption of contaminants which could influence the transport of chemicals and metals of concern (Bastian 2009).
This is referred to as the “Trojan horse” carrier concept (e.g. arsenic adsorbed to the surface of a nanomaterial travelling across a cell membrane; Shipley 2009).
3.c. Hazard Assessment
51. Hazard Assessment for Classification and Labelling (C&L). It is noted that irrespective of the exposure, a hazard assessment is needed for the purposes of C&L of any substance, including manufactured nanomaterials.
52. Identification of the Toxic Principle13. The SCENIHR 2006 considered the identification of the toxic principle of a given nanomaterial a critical issue to be resolved initially in hazard assessment. The hazard may be due principally to, for example:
• the toxicological properties of the chemical(s) that comprise the core of the nanoparticle, or the influence of functionalisation of the nanoparticle surface;
• the much greater relative surface area of the nanoform and, consequently, the greater potential reactivity;
• the potential, due to the enhanced surface area and possible surface reactivity, for other chemicals of concern to be absorbed onto the nanoparticles; and/or
• contaminants and/or by-products related to the nanoparticle production (e.g., metal catalysts).
• If contaminants or by-products related to the nanoparticle production (e.g., metal catalysts) are responsible for the hazard, this may also affect the results of risk assessment and management (Liu et al. 2008).
53. Dose Metrics. For nanomaterials the actual metric that best describes the observed effects in test organisms or environmental fate and distribution may not be mass-based, usually expressed as mg/kg body weight or mg/L (or mg/m3). There are indications that, for example, the number of nanoparticles, the surface area, or another metric can be in some cases a better metric to relate dose to the observed fate, behaviour, and effects of a specific nanomaterial relative these effects across a range of particle sizes (also rf. to chapter 2 and Aitken 2011, Hankin 2011). Knowledge about the mechanisms underlying the observed effect (but also determining fate) would be required to make a decision on the scientifically most appropriate dose metrics on a case-by-case basis or for defined groups on nanomaterials. Using another metric, however, will need further discussion, and might also have major consequences for the international Mutual Data Acceptance as well as for most legislations. Altering the metrics for hazard would require also using consistent units for exposure and risk estimation. This includes classification and labelling, where most hazards of a substance are related to mass concentration.
54. Material Heterogeneity and Batch-to-Batch Variation. Depending on the outcome of the
“Nanomaterial Identification” step (see above), there may be substantial variation in properties between samples of the “same” material from producer to producer and/or batch to batch14. For carbon nanotubes (CNTs), for example, especially multi-walled CNTs, variation in length, metal content, aggregation and surface chemistry of the produced material is known and expected to influence measured toxicity (Johnston 2010). Heterogeneity in the degree of surface modification and/or aggregation was reported to
13 Toxic Principle: Describes the constituent or the substructure of a given material that is responsible for the toxic effects of that material (e.g. an impurity, an aspect ratio, a surface charge etc.)
14 In principle, this consideration is not limited to nanomaterials and applies to any other form of chemical substances, but the typical spectrum of properties affected would be expected to be different (see also chapter 2).
influence the activity of fullerenes (Chae 2010) and other nanoscale materials. Such variability may cause (quantitative) differences in toxicological effects and thus affect the outcome of hazard characterisation.
55. Relevant Nanomaterial (Sub) Species. The specific form(s) or (sub)species of a heterogenous nanomaterial to which humans or the environment may finally be exposed, as well as the specific activities of this particular fraction are largely unknown. This situation is further complicated by the fact that during their life-cycle, nanomaterials can transform from one form to another (e.g. coated – uncoated, oxidised - reduced), agglomerate or aggregate, and dissolve in part or completely. Therefore, information on the state of the nanomaterial in situ and the specific form that causes the observed effect could potentially reduce the degree of uncertainty.
56. Linking Material Properties to ADME (Absorption, Distribution, Metabolism and Elimination) and Toxic Effects. In view of the diversity of nanomaterials also within one “material class”, there is an urgent need for valid approaches to categorise or otherwise group nanomaterials in order to allow read-across or bridging of data for assessment (and decision making). Development of an understanding how material properties are linked to ADME and toxicity would assist in building of categories and enable QSAR approaches.
57. Definition of Biologically Relevant Properties. A well known example of how abiotic and biotic factors influence bioavailability, bioaccumulation and toxicity of nanomaterials is that of asbestos fibres15. Here, it is accepted that a combination of the aspect ratio of the fibre (or shape) i.e. length and width, and the durability or biopersistence of the nanofibre, in the context of the physiological response in the airways and the macrophages in the lung (i.e. clearance), are the critical determinants of subsequent toxicity and pathology. This demonstrates the need to acknowledge and understand such complex interactions when predicting toxicity and pathogenicity for nanomaterials when exposure occurs in natural settings. Without such knowledge, the descriptors chosen for substance identification may be unsuitable, leading to inclusion of less or non-hazardous material in a high hazard category and vice versa.
58. Biological Relevance of Testing Conditions. Another consequence of such complex interactions between nanomaterials and abiotic and biotic factors is that exposure to single nanomaterials following some of the recommended standard test protocols (e.g. a daphnid or fish test using standardised de-ionised water) may have only limited relevance when compared with the natural environment in which exposure occurs. There, abiotic and biotic factors can greatly change the structure, form, behaviour and fate of nanomaterials and may thereby influence their bioavailability and toxicity to a larger extent and through other reactions than known or expected for conventional (non-nano)materials. Relevance is likely to vary with both the method chosen (in vitro / in vivo) and exposure route (e.g. water, air).
59. Assessment of the Quantitative Relevance of Testing Conditions. The implications of interacting factors such as, for example, dispersion media and protocol and their unknown relevance are that a considerable measure of uncertainty is introduced to the calculation of a Lowest or No Observed Effects Concentration (LOAEC/NOAEC) when using some of the current standard tests employed for chemicals. This is an issue that should be further considered by SG4 in its review of OECD test methods.
59. Assessment of the Quantitative Relevance of Testing Conditions. The implications of interacting factors such as, for example, dispersion media and protocol and their unknown relevance are that a considerable measure of uncertainty is introduced to the calculation of a Lowest or No Observed Effects Concentration (LOAEC/NOAEC) when using some of the current standard tests employed for chemicals. This is an issue that should be further considered by SG4 in its review of OECD test methods.